Artificial Intelligence and Common Sense: The Shady Future of AI

Abstract

Artificial Intelligence (AI) revolution has recorded more impact than any other revolutions in human history. AI systems have been more far ahead with the recent advances particularity with deep learning that attained state-of-the-art results in almost all Machine Learning (ML) tasks. However, AI systems are venerable but are highly vulnerable which demands timely human intervention. Further, AI systems can easily be tweaked and misguided to produce misleading results that are far from the reality. It is high time to address this susceptibility of AI since reliance on AI systems is keep growing exponentially. AI lacks the key aspect of human intelligence—common sense that guides humans to take better action and decision based on consequences and makes humans more adaptable. This missing aspect of AI inspired the researchers to work toward Artificial general Intelligence (AGI). AGI research involves developing AI systems with human-like consequential and conscious learning. This paper presents the theoretical and practical vulnerability of AI through literature, examples, experiments. The literature and examples concentrates on famous and popular AI systems like deep learning, Google Translate, visual cognition, etc. The experiments are carried out using two datasets; a gene expression dataset for prediction and image dataset for object detection and scene recognition. The experiment results reassert the weakness of AI and the requirement of AGI.